Establishment and efficiency analysis of the risk prediction model predicting the likelihood of bronchopulmonary dysplasia in preterm infants: A retrospective observational study

Abstract

Background Clinically useful and adequately validated predictive models of bronchopulmonary dysplasia (BPD) available at or soon after birth, especially for the Chinese population are few. This study aims to develop and validate a predictive model for the identification of BPD at early stage and the generation of a nomogram for clinical use.Methods This retrospective study developed a well-validated predictive model of BPD among preterm infants with gestational age (GA) of less than 32 weeks by logistic regression and generated a nomogram.Results This model comprised six variables, i.e., GA, birthweight, neonatal asphyxia, invasive ventilation, neonatal respiratory distress syndrome (NRDS), and patent ductus arteriosus (PDA). The regression equation was Logit P = (15.209 – 0.076 × GA – 1.001 × birthweight + 0.557 × neonatal asphyxia + 1.083 × invasive ventilation + 0.622 × NRDS + 0.656 × PDA). The area under the receiver operating characteristic (ROC) curve of this model was 0.856 (95% CI: 0.812–0.901), with sensitivity of 81.6% and specificity of 76.8%. The verification of this predictive model showed a sensitivity of 93.5% and specificity of 76.5%, demonstrating that the effects of this model were satisfactory.Conclusions This risk prediction model had a good predictive effect for the risk of BPD in premature infants with GA less than 32 weeks and could furnish evidence for preventive treatment and intervention.